1: \section{Analysis updates}\label{sec:finaltouches}

2: 

3: The updates made to the analysis of Ref.~\cite{aleph00}, mentioned in

4: Section~\ref{sec:introduction}, are described in detail in the

5: following subsections. The effect of each of these updates on the

6: significance of the observed excess~\cite{aleph00} is displayed in

7: Table~\ref{tab:effects}. The properties of the most significant

8: four-jet candidates, after all the analysis updates are taken into

9: account, are listed in Table~\ref{tab:4jetcand}.

10: 

11: \begin{table}[thb]

12: \begin{center}

13: \caption{\small 

14: The successive effect of the analysis changes on the maximum

15: significance of the observed excess, for the two alternative analysis

16: streams.

17: \label{tab:effects}}

18: \vspace{0.3cm}

19: \begin{tabular}{|lcc|}

20: \hline

21: Update                  & Cut stream   & NN stream \\

22: \hline

23: \hline

24: Significance~\cite{aleph00} & $\phantom{+}3.06\sigma$   

25:                             & $\phantom{+}2.96\sigma$ \\

26: \hline

27: Final processing        & $+0.21\sigma$ & $-0.14\sigma$\\

28: %LEP $\roots$           & $+0.01\sigma$ & $-0.01\sigma$\\

29: LEP $\roots$            & $-$           & $-$ \\

30: Additional simulated & & \\

31: ~event samples         & $-0.36\sigma$ & $-0.14\sigma$\\

32: %~event samples         & $-0.37\sigma$ & $-0.17\sigma$\\

33: Beam-background         & $+0.13\sigma$ & $+0.14\sigma$\\

34: \hline                                          

35: Final significance      & $\phantom{+}3.04\sigma$ & $\phantom{+}2.82\sigma$\\

36: \hline

37: \end{tabular}

38: \end{center}

39: \end{table}

40: %

41: \vspace{-1.0cm}

42: %

43: \begin{table}[thb]

44: \begin{center}

45: \caption{\small

46: Details of the five four-jet candidates selected with an event weight

47: greater than 0.3 at $\mh=115\,\Gcs$ in either the NN or cut

48: streams. Jets~3 and~4 are the Higgs boson jets. The weight

49: $w=\ln\left(1+{s}{f_s}/{b}{f_b}\right)$ of the candidates in each stream is also

50: given. For candidate $e$, the jet pairing shown is only selected in

51: the cut stream.

52: \label{tab:4jetcand}}

53: \vspace{0.3cm}

54: %\begin{tabular}{|c|c|c|c|c|c|c|c|c|}

55: %\hline

56: %Candidate & $\mrec$ & $m_{12}$ & $m_{34}$ &

57: %\multicolumn{4}{c|}{b tagging} & Four-jet \\

58: %\cline{5-8}

59: %(Run/Event) & (\Gcs) & (\Gcs)& (\Gcs) & Jet 1 & Jet 2 & Jet 3 & Jet 4 & NN \\

60: %\hline

61: %\hline

62: %$a$(56698/7455)& 109.9 & 95.7 & 105.4 & 0.999 & 0.831 & 0.999 & 0.197 & 0.999 \\

63: %$b$(56065/3253)& 114.4 & 95.0 & 110.6 & 0.996 & 0.663 & 1.000 & 0.996 & 0.997 \\

64: %$c$(54698/4881)& 114.1 & 101.1& 104.3 & 0.124 & 0.012 & 0.998 & 0.999 & 0.997 \\

65: %$d$(56366/0955)& 114.4 & 78.6 & 127.0 & 0.201 & 0.051 & 0.998 & 0.956 & 0.933 \\

66: % following combination only selected by cut stream

67: %$e$(55982/6125)& 114.4 & 80.0 & 125.6 & 0.071 & 0.306 & 0.449 & 0.998 & 0.687 \\

68: %\hline

69: %\end{tabular}

70: \begin{tabular}{|c|c|c|c|c|c|c|c|c|}

71: \hline

72: Candidate & $\mrec$ & 

73: \multicolumn{4}{c|}{b tagging} & Four-jet & 

74: $w_{\mathrm{NN}}$ & $w_{\mathrm{cut}}$\\

75: \cline{3-6}

76: (Run/Event) & (\Gcs) & Jet 1 & Jet 2 & Jet 3 & Jet 4 & NN & & \\

77: \hline

78: \hline

79: $a$ (56698/7455)& 109.9 & 0.999 & 0.831 & 0.999 & 0.197 & 0.999 & 0.59 & 0.25 \\

80: $b$ (56065/3253)& 114.4 & 0.996 & 0.663 & 1.000 & 0.996 & 0.997 & 1.44 & 0.81 \\

81: $c$ (54698/4881)& 114.1 & 0.124 & 0.012 & 0.998 & 0.999 & 0.997 & 1.76 & 0.61 \\

82: $d$ (56366/0955)& 114.4 & 0.201 & 0.051 & 0.998 & 0.956 & 0.933 & 0.41 & 0.62 \\

83: % following comb only selected by cut stream 

84: % (a different comb. is now selected by NN, w/ mrec=100.2 GeV)

85: $e$ (55982/6125)& 114.4 & 0.071 & 0.306 & 0.449 & 0.998 & 0.687 &  --  & 0.63 \\

86: \hline

87: \end{tabular}

88: \end{center}

89: \end{table}

90: %

91: \vspace{-1.0cm}

92: %

93: 

94: \subsection{Final processing}

95: 

96:  The data were reprocessed with the final detector calibration and

97:  alignment constants.  This reprocessing resulted in the recovery of

98:  1\,${\rm pb}^{-1}$ of data.  The total integrated luminosity for the

99:  year 2000 is ${\cal{L}}=217.2\,\invpb$.

100: 

101:  The reprocessing can change by small amounts the value of measured

102:  event properties such as the reconstructed Higgs boson mass or the

103:  b-tagging probabilities. Events close to some of the selection cuts

104:  may therefore move into or out of the selected sample.

105: %

106:  About 95\% of the data events selected previously were also selected

107:  after the final processing. More specifically, the most signal-like

108:  events, i.e., those with a large contribution to the log-likelihood

109:  ratio $-2\ln\,Q$, are still selected after the final processing.

110: 

111:  In the cut-based four-jet channel, a new event is selected with a

112:  reconstructed Higgs boson mass of $111.8\,\Gcs$. Prior to the

113:  reprocessing, this event narrowly failed one of the b-tagging

114:  cuts. The two Higgs-candidate jets have b-tagging values of 0.870 and

115:  0.965, whereas the Z-candidate jets have b-tag values of 0.096 and

116:  0.277. (The output of the neural network b-tagging algorithm ranges

117:  from 0. for light-quark jets, to 1. for b-quark jets.) The missing

118:  energy of the event is 70 GeV and the total missing momentum is below

119:  $10\,\Gc$. A probable explanation for the large missing energy and low

120:  missing momentum is that two energetic neutrinos were produced almost

121:  back to back by two b-quark semileptonic decays. Indeed, in one of

122:  the b-tagged jets, an identified muon has a momentum of 1.7\,GeV/$c$

123:  transverse to the jet axis and is therefore consistent with a

124:  semileptonic decay of a b hadron. Another low-momentum muon is

125:  observed opposite to this jet, which further substantiates this

126:  hypothesis.

127:  

128:  In the cut stream, where only the reconstructed mass information is

129:  used as a discriminant, the event is assigned a weight

130:  $\ln(1+{s}{f_s}/{b}{f_b})=0.27$ at $\mh=115\,\Gcs$. In the NN stream,

131:  this event was already selected prior to the final processing. The

132:  event has a NN output of 0.90, and is therefore assigned a relatively

133:  low weight compared to the most significant

134:  candidates~\cite{aleph00}.

135: 

136: \subsection{LEP centre-of-mass energy}

137: 

138:  In the most recently available determination~\cite{lepecal}, the

139:  centre-of-mass energies are, on average, smaller than those used in

140:  Ref.~\cite{aleph00} by $\sim 140$\,MeV. When this effect is taken

141:  into account, the reconstructed Higgs boson masses of the candidate

142:  events are reduced by the same amount, and the number of signal

143:  events expected to be produced decreases from $10.1$ to $9.5$ for

144:  $\mh=115\,\Gcs$. The impact on the observed significance is

145:  negligible.

146: 

147:  \subsection{Additional simulated event samples}

148: 

149:  In order to further reduce the statistical uncertainty in the event

150:  selection efficiencies and in the pdf's, significantly larger event

151:  samples were generated. In particular, additional simulated

152:  background samples for the $\ee\r\bb(\gamma), \cc(\gamma), \WW$ and

153:  $\ZZ$ processes were generated at centre-of-mass energies $\sqrt{s}=$

154:  206.0, 206.7 and 207.0 GeV. The existing signal samples were also

155:  supplemented with samples of \ee\r\h\qq\ and \h\nunu\ events at

156:  $\sqrt{s}=206.7$ GeV. While most of these additional samples were

157:  used in the NN stream for the preliminary results~\cite{aleph00},

158:  they have only been included in the cut stream for this letter.

159: 

160: \subsection{Control of beam-induced backgrounds}\label{sec:beam-bgds}

161: 

162: % Make some statement as to why this is mostly a problem in the four-jet analysis

163: % and not in the Hll/tau channels. (Hvv is naturally protected by E12 cut)

164: 

165:  In one of the most significant four-jet events, called ``candidate

166:  {\it b}'' in Table~\ref{tab:4jetcand}, a 22\,GeV energy deposit is

167:  observed at small polar angle, in the plane of the collider.

168:  

169: %

170:  This deposit does not fit the hypothesis that it is part of the

171:  event. The total measured energy is considerably larger than

172:  $\sqrt{s}$ and the total measured momentum is aligned with that

173:  of the deposit.

174: %

175:  A reasonable kinematic fit quality is obtained only if this deposit

176:  is assumed to be extraneous to the event, i.e., produced by a

177:  beam-induced background particle.

178: 

179:  It is indeed possible to observe large energy clusters from this

180:  background source. For example, in 0.89\% (0.48\%) of events

181:  triggered at random beam crossings, a deposit of energy in excess of

182:  3 (10) GeV is observed. The angular position of the most energetic

183:  cluster observed within 12$^\circ$ of the beam axis, in the

184:  randomly-triggered event sample, is shown in

185:  Fig.~\ref{fig:beam-bgd}a. The overwhelming majority of the

186:  beam-induced background particles are at very small polar angles and

187:  in the plane of the collider.

188: 

189: \begin{figure}[htb]

190: \begin{picture}(170,80)

191: \put(0,0){\epsfig{file=fig1a,width=7.6cm}} 

192: \put(08,50){\Large \bf (a)}

193: \put(07,76){\Large ALEPH}

194: \put(85,0){\epsfig{file=fig1b,width=7.6cm}}

195: \put(96,50){\Large \bf (b)}

196: \put(92,76){\Large ALEPH}

197: \end{picture}

198: \caption{\small 

199: %(a) The energy distribution of the most energetic cluster observed in

200: %the detector within $12^\circ$ of the beam axis, in events triggered

201: %at random beam crossings; 

202: (a) The angular distribution of the most energetic cluster observed

203: within 12$^\circ$ of the beam axis, in a sample of events from

204: randomly-triggered beam crossings. Only clusters with $E>$\,3~GeV are

205: shown. The candidate event {\it b} is indicated by the cross in the

206: upper-left corner. The plane of the collider is defined by $\phi=0$

207: and $\pi$. (b) The distribution of $R$ for the events selected by the

208: cut-based four-jet search, for the expected SM background with (shaded

209: histogram) and without (hatched histogram) contamination from

210: beam-related background, and for the data (dots with error bars).

211: \label{fig:beam-bgd}}

212: \end{figure}

213: %

214:  As this type of background is not simulated, a procedure to

215:  identify and remove beam-background clusters had to be developed.

216: %

217:  The most energetic cluster with energy greater than 3 GeV, \mbox{$|\cos

218:  \theta\,|>0.998$} and which is isolated by at least \mbox{8$^\circ$}

219:  with respect to any other particle in the event, is fitted to each of

220:  the following three hypotheses.

221: 

222: \begin{itemize}

223: \item The identified cluster is part of the event.~In this case, 

224: the identified cluster is assigned to one of the four jets by the jet

225: clustering procedure.~The jets are subsequently fitted to the total

226: energy- and momentum-conservation constraints.

227: 

228: \item The identified cluster is, more specifically, 

229: assumed to be an ISR photon.~In this case, the rest of the event is

230: forced to form four jets.~These jets are fitted to the total energy-

231: and momentum-conservation constraints, modified to account for the

232: momentum imbalance caused by the hypothetical ISR photon.

233: 

234: \item The identified cluster is assumed to originate from a 

235: beam-induced background particle.~In this case too, the rest of the

236: event is forced to form four jets.~These jets are subsequently fitted

237: to the total energy- and momentum-conservation constraints.

238: \end{itemize}

239: %

240:  The $\chi^2$ values of these fits are henceforth designated

241:  $\chi_{\mathrm{norm}}^2, \chi_{\mathrm{ISR}}^2$ and

242:  $\chi_{\mathrm{beam}}^2$, respectively. The ratio

243: 

244: \[ R = \frac{\min(\chi_{\mathrm{norm}}^2,\chi_{\mathrm{ISR}}^2)}

245: {\chi_{\mathrm{beam}}^2}\] is expected to be larger for events

246: containing a beam-background particle. The distribution of $R$ for the

247: total expected SM background is shown in Fig.~\ref{fig:beam-bgd}b,

248: before and after it is ``contaminated'' with beam-background clusters

249: obtained from a sample of randomly-triggered events. Events in which

250: no energetic, isolated, small-angle cluster is found are assigned

251: $R=0$\,. Events with $R>2.$ are tagged as containing a beam-background

252: particle and the identified cluster of energy is removed from the

253: event prior to jet clustering and kinematic fitting. The remaining

254: events are treated according to the first hypothesis.

255: 

256:  The efficiency of the beam-background cleaning procedure,

257:  determined by running the algorithm on a contaminated background

258:  sample, is 28\% (50\%) for events with energy deposits in excess of

259:  3 GeV (10 GeV). The purity of the identification procedure is close

260:  to 100\%.

261: 

262: % Now give details about CLEAN vs DIRTY+CLEANED MC:

263:  At the final selection level 1.2\% of the simulated events are

264:  affected (i.e., newly selected, no longer selected, or with an

265:  $M_{\rm rec}$ value changed by at least 1\,GeV/$c^2$) by the

266:  contamination. This fraction is reduced to 0.4\% after the cleaning

267:  procedure is applied. The corresponding changes to the selection

268:  efficiencies are statistically insignificant and the changes in the

269:  pdf's imperceptible.

270: 

271:  When applied to the data, the cleaning algorithm identifies only one

272:  event (candidate {\it b}) as containing a beam-induced energy

273:  deposit. The deposit is therefore ignored in the analysis of this

274:  event, and the reconstructed Higgs mass (neural network output)

275:  changes from $112.8\,\Gcs$ (0.996) to $114.4\,\Gcs$ (0.997).

276: 

277:  

278: 

279: 

280: %%% Local Variables: 

281: %%% mode: latex

282: %%% TeX-master: t

283: %%% End: 

284: